Flow Orientation as a Critical Parameter in Nanoscale Membrane Filtration for Optimising Small Extracellular Vesicle Isolation

Abstract

Membrane filtration of biological nanoparticles typically employs tangential or perpendicular flow geometries, each presenting inherent performance trade-offs between membrane flux and fouling. While these configurations involve established limitations in transmembrane pressure, fouling, and particle damage, how intermediate inlet angles quantitatively modulate these effects at the microscale remains unexplored. Here, we combine computational fluid dynamics with experimental validation to systematically investigate how inlet microchannel orientation relative to a nanoporous alumina membrane (0°, 30°, 60°, and 90°) governs the separation performance of small extracellular vesicles (sEVs) – sub-200 nm particles that serve as promising liquid biopsy targets for ovarian cancer detection. Simulations show that intermediate flow angles (30° and 60°) produce filtration fluxes between those of perpendicular and tangential configurations, reflecting modified near-membrane hydrodynamics, while potentially improving filtrate composition. Using OVCAR3 ovarian cancer cell-derived sEVs, we demonstrate that a 60° inlet angle generates a distinct hydrodynamic regime favourable for sEV membrane-based filtration. Devices with inlet orientation at 60° i) suppresses contaminants of >100 nm by 3-fold compared to perpendicular flow, ii) maintains comparable sEV recovery compared to tangential flow (p < 0.05), and iii) enhances biomarker detectability, with CD9 increasing up to 6-fold and CA125 up to 2.5-fold relative to conventional geometries. Together, these results identify flow orientation as a key design parameter in nanoscale membrane bioseparation, showing that intermediate geometries create distinct transport behaviours. They also provide experimentally validated foundations for advancing nanofiltration-enabled sEV isolation, directly relevant to improving ovarian cancer diagnostics given the value of sEV‑associated CA125 as an early‑stage biomarker.